Current header compression is used on point-to-point (PPP) networks. Such compression reduces TCP/IP overhead from about 40 bytes to 4 bytes. In typical packet data networks, packets are often dropped at a selection distribution unit, due to the bottleneck provided on the over the air radio link. Dropped packets cause a lack of synchronization in the compression mechanism and lead to significant degradation of TCP throughput.
A common method of header compression is referred to as Van Jacobson header compression. Van Jacobson header compression works, in a general sense, based on delta-encoding between two end points in a PPP link. Each packet is sent with a sequence number. Rather than sending the sequence number in the header, the header includes a field that indicates the change in sequence number from the previous packet. This allows for compressed headers, due to the lessening of the data field required for the delta.
One problem associated with such a compression scheme is that if a packet gets dropped between end points in the PPP network, the delta information in the header relating to the sequence number is no longer valid. Successive packets are dropped because of a mismatch between the sequence number and the checksum. As a result, multiple packets are dropped, and the system does not right itself until the sender times out due to not receiving an acknowledgement from the receiver in time. Consequently, multiple packets are lost, and additional time is required to get the packet network re-synchronized.
Thus, a need exists for a method for retransmitting a data packet in a packet network which provides for increased reliability and fewer dropped packets.